Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Parallel connection of multiple half-bridge for high current motor applications

Status
Not open for further replies.

H2M

Member level 1
Joined
Aug 11, 2016
Messages
36
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,286
Activity points
1,624
Hi, I want to build a 15Kw DC motor driver. Is it possible to design five 3Kw half-bridge motor driver and connect them in parallel?
 

Hi,

generally it depends on drive mode and hardware if it is theoretically possible at all.
But then the next problem arises: How can you be sure that every half bridge drives exactly 1/5 of the total current?

Btw: If you design a halfbridge: you don´t design it for "power" you design it for voltage and current.

Klaus
 

Thanks, Klaus,
How can I pass 300Amps through Mosfet? The only way is to parallel them, I did it before for a 1Kw motor driver but I used a single half-bridge driver and two parallel MOSFETs in each side of half-bridge. If I want to parallel more than two MOSFETs to make a 15Kw motor driver, I will have a limitation for source/sink current of the gate driver. Do you have any suggestion?
 

Unitrode, now part of Texas Instruments, had a very nice app note detailing methods for paralleling power supplies for higher power, with all supplies sharing the current equally.

The most difficult part is making sure that the current sharing behaves correctly during transient conditions.

We are talking regular DC motors, correct? Not BLDCs.
 

Yes, I want to make a speed controller for a 15Kw Brushed DC motor. In DRV8412 datasheet(Texas Instruments PWM DC motor driver) they connect an inductor in series with half-bridge outputs and then parallel them. Can I use this method?
DRV8412.PNG
 

Hi,

I'd not build individual halbridges...
I'd build one halfbridge, with paralleled transistors.

Klaus
 

It's reasonable to implement the motor driver as multi phase converter to reduce input and output ripple current.
 

Hi,

I see the big benefit of multiphase when you need (low ripple) DC output. There you need inductances anyways.

With motor control there is no need to have DC output, a motor will run with PWM.
Thus the drawback is that you additionally need big (size, costly) inductances with polyphase topology.
But for sure the polyphase will reduce EMI noise..

Klaus
 

Hi,

A general idea for paralleling power units:
* make one unit as the "master"
* all other units are current controlled slaves (to deliver the same current as the master)

You need to take care about system stability.

Klaus
 

In this module, they used a single MOSFET in each side! Also, they mentioned to "Ease of paralleling" in their features without any additional elements. If I directly parallel several modules I must consider enough dead time to avoid shoot-through between modules. Also by increasing the dead time, the total power dissipation will be increased.
 

Hi, I want to build a 15Kw DC motor driver. Is it possible to design five 3Kw half-bridge motor driver and connect them in parallel?

or use something like this ...

Power MOSFET Module 150 Volt / 300 Amp continuous @ 90° C
https://www.vishay.com/docs/96060/vs-fc420sa15.pdf

If you decide to use 5 ( or x number ) MOSFETS in parallel then make sure that the Peak Amps Rating of any one MOSFET can temporarily carry 100% FULL LOAD AMPS.
When turning ON ... the first MOSFET to turn ON will carry a huge load until every other MOSFET turns ON.
When turning OFF ... the last MOSFET to turn OFF will carry a huge load until it finally turns OFF.

Please do not try to convince anybody that you can turn every MOSFET on or off at the exact same time.
Think about what happens when only one of your 3 KW MOSFET Half-Bridges turns ON, while the other 4 bridges are still OFF ...
How long will it last before it releases the magic smoke ?
 
Last edited:

Hi,

In this module, they used a single MOSFET in each side
Are you talking about the CAS300 modules of post#9.
Is this written in the datasheet? I can´t find this information. How do you know?

Also, they mentioned to "Ease of paralleling" in their features
Whatever that means.
Paralleling MOSFETs is always more easy then paralleling BJTs, because of their temperature behaviour.

without any additional elements
How do you know?

You may be corrrect or not - I`m not able to validate it.
***
... I assume there are multiple dice in one module. (I don´t know for sure)
With multiple dice on one module it´s more easy to ensure equal temperatures, because they are all mounted on the same heat spreading plate....with identical production parameters.
Even the dice may be better electrically matched when they come from one production lot.

Did you notice the price? more than 2500€ per 5 pieces at Farnell.
For this money you could buy a lot of MOSFETS and parallel them with

I did a couple of high current regulation applications .. (up to 6000 A RMS).
300A with such modules is manageable, but for sure you need to take care with wiring, driving the gates and temperatures.

Klaus

Still we have no voltage and current value from your side.
In post#5 you talk about 15kW and show a 50V device .... this means at least 300A.

In post#9 there is a 1200V module ... here you need just 12.5A to get 15kW.
For sure 12.5A is less critical than 300A.

- - - Updated - - -

Hi,

If you decide to use 5 ( or x number ) MOSFETS in parallel then make sure that the Peak Amps Rating of any one MOSFET can temporarily carry 100% FULL LOAD AMPS.
When turning ON ... the first MOSFET to turn ON will carry a huge load until every other MOSFET turns ON.
When turning OFF ... the last MOSFET to turn OFF will carry a huge load until it finally turns OFF.
True.
And even worse when the "slow turn off" one keeps the voltage while the others even faster switch OFF because they don´t see the miller plateau.
This means the faster ones have much less switching loss than the slow one.

Klaus
 

If you decide to use 5 ( or x number ) MOSFETS in parallel then make sure that the Peak Amps Rating of any one MOSFET can temporarily carry 100% FULL LOAD AMPS.
When turning ON ... the first MOSFET to turn ON will carry a huge load until every other MOSFET turns ON.
When turning OFF ... the last MOSFET to turn OFF will carry a huge load until it finally turns OFF.
In typical switching applications, source inductance is the limiting factor for switching speed and will almost perfectly balance the current rise and fall timing of parallel connected MOSFETs.
 

MOSFET Turn-On is not an event ... it can be described in a 4 step process that takes time

Typically, the MOSFET turn-on process can be described in 4 steps ...
Process #1 ... from T0 until T1 = Current begins to flow from Gate Driver to the MOSFET Gate
Process #2 ... from T1 until T2 = Gate Threshold voltage achieved and current begins to flow
Process #3 ... from T2 until T3 = Miller Plateau in Gate Voltage
Process #4 ... from T3 until T4 = The Rds has finally reached its minimum value

How long each one of these processes takes, is determined by:
a) Variances in the Gate Drivers
b) Variances in Board / Trace Layout
c) Variances in Gate Capacitance
d) Variances in Gate Threshold Voltage
e) Variances in Gain Factor
f) Variances in Branch Inductance
g) Variances in Source Inductance
e) Variances in Junction Temperature

Both Vishay and IRF show that "Turn-On" of Parallel MOSFETs does not occur at the same time.
Also, the final Rds is not equal, which leads to unequal current flow
Unequal current flow leads to a higher junction temperature in one MOSFET.
Unequal junction temperatures make the MOSFET's behave differently.


An example from IRF
=============
One MOSFET at max amps ...
MOSFET #1 = 800 u joules per On/Off switching event

Two Parallel MOSFET's, with twice the load ...
MOSFET #1 = 1200 u joules per On/Off switching event <<< Way too HOT !!!
MOSFET #2 = 400 u joules per On/Off switching event


Thinking you can just parallel two MOSFET's and then control twice the load, can cause one MOSFET to overheat & run outside of its SOA.


To make Parallel MOSET's a success ...
a) All of the components must be very well matched
b) Differences must be compensated for
c) The MOSFET's should be over-sized, for the worst case scenario
d) Cooling should be over-sized, for the worst case scenario.

Five (5) x 3 kW Half-Bridge Drivers might not be "safe" with a 15 kW motor.
Units #2 thru #5 have a slightly higher resistance and carry only 90% of their fair share vs Unit #1 ...

3 kW Unit #1 @ 28% x 15KW = 4.2 kW <<< Yikes !!! ( a perfect example of why / how parallel mosfets eventually burn-up )
3 kW Unit #2 @ 18% x 15KW = 2.7 kW
3 kW Unit #3 @ 18% x 15KW = 2.7 kW
3 kW Unit #4 @ 18% x 15KW = 2.7 kW
3 kW Unit #5 @ 18% x 15KW = 2.7 kW

Five (5) x 3 kW Half-Bridge Drivers might be safe with a 10 kW motor ...
Again, Units #2 thru #5 have a slightly higher resistance and carry only 90% of their fair share vs Unit #1 ...

3 kW Unit #1 @ 28% x 10 kW = 2.8 kW = OK
3 kW Unit #2 @ 18% x 10 kW = 1.8 kW = OK
3 kW Unit #3 @ 18% x 10 kW = 1.8 kW = OK
3 kW Unit #4 @ 18% x 10 kW = 1.8 kW = OK
3 kW Unit #5 @ 18% x 10 kW = 1.8 kW = OK

It is all about the variances.
 
Last edited:

But I have to agree with Klaus. The OP has not told us yet the motor's plate voltage.

Without this information, any recommendations we will give may not apply to his project, and we are blindly providing recommendations.

For instance, the DRV8412 shown is only 50 volts max. However a 15kW motor is a 20 HP motor, and a quick web search indicates that these power levels require at least 240 armature volts.
 

Thanks, for your nice tips.
In the CAS300 datasheet, they draw a single MOSFET half-bridge in the body of the module, Also in the last page of datasheet. First I thought they used one MOSFET in each side of half-bridge. But when I searched more I realized that they use these MOSFETs:
https://eu.mouser.com/new/wolfspeed/cree-c2m/
All of these MOSFETs have an RDS between 25mohm to 1ohm:
RDS.PNG
But the RDS of CAS300 module is 5mohm, so they are used parallel MOSFETs.
In some Application note, I have seen directly connection of half-bridge module:
https://www.st.com/content/ccc/reso...df/jcr:content/translations/en.DM00310534.pdf
They propose directly parallel connection:
Parallel Conection.PNG
But from your tips, I will parallel high current MOSFETs with single high source/sink current gate driver instead of paralleling multiple half-bridge modules with individual gate dirver.
I am working on a 48 volts permanent magnet DC motor with 15Kw nominal power.

- - - Updated - - -

There is attached a picture of my motor characteristics:
12KW-48V.gif
Also, I attached the no-load test of this motor in this movie:
https://www.youtube.com/watch?v=8EGqecPmtRk
 

Attachments

  • 12KW-48V.gif
    12KW-48V.gif
    156.2 KB · Views: 105
Last edited:

Cree CAS300 modules are of course using multiple dies, like other high current MOSFET or IGBT modules. In contrast to parallel connection of arbitrarily selected discrete transistors, you can expect some Vgs,th matching.

I will parallel high current MOSFETs with single high source/sink current gate driver instead of paralleling multiple half-bridge modules with individual gate driver.
Delay skew is the parameter of interest. On the other hand, it's required to place the final driver near to the transistor. A possible solution may be a common driver with individual boosters, e.g. complementary common collector stages.

We don't got yet an idea of the intended switching frequency.
 

Hi,
I need to parallel about 8 MOSFETs on each side of my Half-bridge to make my 15Kw DC motor driver. If I use a common gate driver, I need more source/sink current. In the AN-978, page17 of IRF they propose a push-pull stage in each output of IR2110 to obtain more current:
Push-Pull.PNG
But I have a question, Can bootstrap capacitor obtain this large current?
 

Bootstrap driver has a number of limitations, particularly not providing 100 % duty cycle and requirement of a specific startup sequence. Can't you afford isolated DC/DC converters for a 15 kW motor drive?

Total driver power depends on the gate charge, switching frequency and additional losses. But there's no principle reason why it can't be sourced by a bootstrap circuit, not considering the above mentioned reservations.
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top